Best books for people changing fields?

Too true. Ralph Morrison's book

Grounding and Shielding: Circuits and Interference (Morrison, Ralph. Grounding and Shielding Techniques.) (Hardcover) by Ralph Morrison (Author) Wiley-IEEE Press; 5 edition (March 16, 2007) ISBN-10:

has been through a lot of editions, and now has chapter on printed ciruit layout for high-speed circuits, but it's not a patch on the rest of the book; the intoduction to the original book is brilliant and makes the basic physics thoroughly accessible. I've yet to come across anything equivalent for high speed circuit design.

Howard Johnson's books

High Speed Digital Design: A Handbook of Black Magic (Prentice Hall Modern Semiconductor Design Series' Sub Series: PH Signal Integrity Library) ISBN-10: 0133957241 ISBN-13: 978-0133957242

High Speed Signal Propagation: Advanced Black Magic (Prentice Hall Modern Semiconductor Design Series' Sub Series: PH Signal Integrity Library) ISBN-10: 013084408X ISBN-13: 978-0130844088

both of which I've got - aren't worth reading through, though they are very handy places to find information about solutions to specific problems.

I've also got a copy of the Williams and Taylor book on filter design

Electronic Filter Design Handbook, Fourth Edition (McGraw-Hill Handbooks) (Hardcover) by Arthur Williams (Author), Fred J. Taylor (Author) ISBN-10:

albeit the second edition rather than the fourth edition, and I've recommended it here from time to time.

I've also recommended

"Coaxial AC Bridges" by B P Kibble and G H Rayner, ISBN 0-85274-389-0

which is great if you happen to want an AC bridge circuit with a precision approaching one part in ten million, which few people do, but sadly it is now out of print and tne U.K. National Physical Laboratory has stopped printing it to order (as they were doing until a few years ago).

-- Bill Sloman, Nijmegen

Yes, that's perfect. I didn't realize that you had edited the list for usenet posting.

Thanks. I see.

To cut costs, would 5000 copies be sufficient to justify a soft cover printing instead of hard cover? Have you considered self-publishing? Is the market size for such scientific books limited and static, or would a substantial price drop increase sales? (Note: I'm not familiar with current publishing economics, am not into optics, and do not consider myself a potential buyer. I'm just curious).

Wait, wait! I decided to pass over these assertions about Terman the first time around, since they seemed so off the wall, not to mention irrelevant to the subject under discussion.

But as someone who's been in the heart of Terman's academic and technical milieu for 54 years now, has interacted with innumerable people who were close to Terman (and had some direct comment with him myself), and has a continuing interest in the development and the history of this whole era, I've never heard or read anything to support statements like those above.

That's the black magic part. With many books the reader has to separate the pixie dust from the real stuff :-)

Yep, if you add it in I'd recommend getting an older copy. Hamfests, EBay, thrift store (if you can find one with books), estate sales (look out for one where there's a humongous antenna near the house).

I wish the ARRL would put a few of them from the 70's/80's online.

I didn't intend to diss Terman--I haven't heard bad reports of him either--though all I know about him comes from his books, and having studied in a building named after him. Did the Varians complain in "The Inventor and the Pilot" or something?

Cheers,

Phil Hobbs

Yes. Read the book.

John

The second edition is about a quarter longer than the first--it'll run to just over 800 large quarto pages, and would have been almost 900 except that Wiley changed to a slightly smaller font. The problem is that softcovers of that size won't stand much abuse. I hate it when I get a book that I planned to use for years and it falls apart. There's a fair amount of new material, and some things that were too terse in the first edition have been expanded significantly. The leading one is the extended example in the signal processing chapter (the laser bug zapper), which has nearly doubled in size.

I think it's mainly overhead that keeps the price highish--it probably takes roughly the same amount of design, editing, accounting, and so forth, but there are many fewer copies to amortise over. Thus saving $5 on printing wouldn't make that much difference, I suspect. The way I picked a publisher in the first place was to look at my bookshelf and see which publisher had the best books. Wiley was streets ahead of McGraw-Hill, Academic Press, and Artech House. Prentice-Hall and Addison-Wesley are generally cheaper but that's because they concentrate on more populated fields.

Self-publishing is a real schlepp--AES (Prof. Siegman) went that route for "Lasers", but I don't have the chops for the job (and anyway I'm not a past president of the Optical Society, which probably helps with the distribution issues). When I buy books, I often compare the time I'll save by working smarter vs the time and money investment. The book is sufficiently practical that I think it will wind up being worth it for anybody with electro-optical gizmos to build, unless they're really expert already and don't need it.

Thanks again,

Phil Hobbs

Yes. Morrison is on the list already--good to know about the PC layout section in the newer edition. Since the first edition is 30 or so years old, Morrison is probably either retired or too senior to be doing much bench work these days. When you're writing about this stuff from a distance, you don't have enough grip on the situation to be really practical, as I've discovered over and over. You really have to have tried escaping 500 high speed differential lines from a CPU on a six-layer board to know what PC design is all about. (No, I haven't, but some of my friends have--one product board I've helped with has a

I'm still sort of torn about it--same with "Photonics Rules of Thumb", which I gave a good review on Amazon, but which turns out on closer examination to have a lot of the same issues as HoJo.

I've been trying to dig up a copy of this for ages, but haven't succeeded so far. I'll happily trade a copy of my second edition for one, if anyone is so motivated.

Thanks again,

Phil Hobbs

Those are excellent suggestions, thanks. I'll do that.

Maguire is a simple and engagingly-written book that helps people avoid writing unmaintainable apps. He's talking about C programming primarily, and the book was written before test-driven design became popular, but otherwise it's a great book. It's full of examples of Microsoft screwups, too, which is fun--props to Microsoft Press for letting him put them in.

Thanks again,

Phil Hobbs

There's one for sale on Amazon for $250. The original cover price was $80.

They (UK Nat Phys Lab) must have the rights to it and it should be possible to secure permission for printing, I'd hope. With that, there are a number of one-off publishers that would print it on-demand. At least, that would make it possible for others to own a copy.

I suppose the problem is getting ahold of someone with the authority to write up a permission letter and time required, perhaps, for an attorney (barrister?) to do the "legal diligence" someone there might feel is needed in writing the permission.

Jon

What stuff there is on high-speed pcb layout is contradictory and mostly bad. Pity.

Parts of it are absurd, and the writing can get extremely annoying.

I have PROT, and I agree, it's sloppy like the Black Magic thing.

John

Even better would be a compilation CD-ROM of all the articles. Some were included for a few years, and some for a decade or more, so they could eliminate the duplication and the advertising. If it was OCRd instead of scanned, it would be 95% text files. They could probably squeeze 20 years worth onto a single CD-ROM.

Jeff,

For a broad senior or first-year-grad-level introduction to modern photonics and fibers, Bahaa Saleh and Mel Teich, Fundamentals of Photonics (Wiley 1991), seems to have been well received; and I think a new edition is coming out.

For a general first-year-grad-level optics and EM text, I've always liked J. M. Stone, Radiation and Optics. New York: McGraw-Hill, 1963. Modern approach, MKS units, good blending of physics and classical optics approach and more engineering-oriented EM-theory approach -- and treats some more modern concepts (dipole radiators, inhomogeneous plane waves) much more clearly than older texts.

Handbook of Optical and Laser Scanning, G. B. Marshall, Ed. New York: Marcel Dekker, (2004) seems to be a good practical book (though I can't judge it with much authority, since that's not my field). In any case the opening chapter by Tom Johnston, Jr. and Mike Sasnett on "Characterization of Laser Beams: The M^2 Model" is as good a tutorial as anyone could want on definition and measurement of laser beam quality.

It may be too advanced (and too Massive!) for your purposes, but Foundations of Image Science by Harrison Barrett and Kyle Myers (Wiley,

I believe Bob Boyd's text on Nonlinear Optics (2nd edition, Academic Press, 2003) has been well received -- and a 3rd edition may be coming out.

Where I'd get into serious discussion with you, however, is on the whole question of handbooks and tables, for numerical and physical data.

Let's take data tables first, because that seems an easier topic. Why would anyone, these days, purchase -- or even give shelf space to -- any kind of tables or handbooks of physical or engineering data of almost any kind, with an Internet connection and Google at hand?

My experience is that Google comes out way ahead not just on speed and convenience, but on breadth and completeness as to what's available, whether you're talking about more basic and general stuff (encyclopedia type stuff, which is everywhere) or the most specialized end (where you have a high chance of being led to some insanely specialized and fanatically complete web site on almost any arcane topic you go after).

Now, what about math data and functions, both algorithms and numbers?

It was claimed some years ago that Mathematica 3 (or was it 4?) contained and could instantly deliver every single bit of mathematical knowledge, both numbers and formulas, contained in Abramowitz and Stegun, and then some; and this seemed a pretty accurate clain.

M4 (or was it 5?) then contained every bit of mathematical knowledge in the average university physics library.

Hyperbole aside, at this pint why would anyone purchase or even give shelf space to mathematical tables or handbooks of any kind -- log tables, table of arcane special functions, or handbooks of numerical algorithms -- with Mathematica available on your laptop?

Numerical methods may be in some cases an exception to this. Anyone who wants to _calculate_ something, as as part of a design or analysis or research process, should just do it in Mathematica -- you'll find the algorithm you need all implement and optimized there.

(And I really don't buy the argument that you need to understand numerical algorithms in order to use or apply them. Nice if you have that understanding as part of your cultural background -- but not essential. If you need to code an algorithm as firmware into limited memory in a gadget, that's another matter -- but then maybe you should hire a consultant.)

All of the above advice, however, is unfortunately currently trapped in the quagmire of the latest version of Mathematica, namely M6.

Mathematica 6 is a magnificent piece of software, so far as I can tell from using it (or more accurately, battling with it, in between periods of screaming frustration): all the old Mathematica math and analysis capabilities, incredible new interactive display capabilities, and just incredible blinding speed of execution.

But as a result of whatever unknown insanity prevails at WRI, they've elected to remove or deprecate or change many older useful features and useful ways of interacting with Mathematica (I don't really object to that, actually); add over 1000 new and complex commands and changes in the interface; ***and then decree that there shall be absolutely NO printed or intelligently planned or helpful introductory-user-level documentation for this vast and vastly changed program*** -- only an even more vast, hard to access, and very unhelpfully designed online "documentation" system.

It's now over a year since M6 came out, and there is still not a book, manual, user guide, introduction to M6, or similar -- not a single piece of printed documentation -- available, from WRI or third parties. Unbelievable.

That might explain why my vision appears to be fading. Since it's a book on optics, you might include a credit card size Fresnel lens magnifier.

A friend just happen to be reading over my shoulder. She works for a library and notes that libraries and universities prefer to buy non-periodical books in hard back format for exactly that reason. Now, I understand why so many technical books are available only in hard cover. She also grumbled that some books are arriving in 20lb paper which is affecting their durability.

Some of my friends are into self publishing. It's actually a family business. They have a series of self-help books that they're selling directly. Suffice to say they're always on the road, giving talks, autographs, deliveries, and promotions. They claim it's worth it because the middle man costs are eliminated. However, they were about to go broke about 10 year ago, when internet sales magically saved their business. I suspect you're not interested in doing the required promotion that is required for self-publishing.

What I was thinking (err... guessing) is that an industry specific technical book has a fairly fixed audience. The bulk of the sales will probably go to institutions, libraries, and individuals in the field. How it's sold seems to be irrelevant. If there is only a market for 5000 books, it would be possible to obtain a mailing list of previous purchases and prospective buyers, suitable for selling directly. The catch is that you have to pay for your printing costs up front.

Most people use similar logic when buying books. However, I find myself looking for books and reference material only AFTER I'm deeply mired in difficulties. Mostly, I look for specific examples that can either explain my current predicament, or can be adapted to fix the problem. That makes book purchases more of a impulsive and panic driven exercise, than one that allows careful research, comparisons, and reading reviews in detail. I have several large and expensive technical books, from which I've read and utilized exactly one chapter.

Good suggestion. I have a copy somewhere at work, which I've never used. I'll have another squint at it.

I've ordered a copy from the library.

I'll have a look at it, thanks.

A bit far afield for the electrooptical systems crowd, I think?

Something more recent than Nico Bloembergen would be good, it's true. I'll have a look at it.

Basically for the same reason I write a lot of my own code: it'll still be there next time I want it. I'd far rather look up something in a familiar, well laid out source such as Abramowitz and Stegun than in somebody's help system. (I *hate* help systems, especially Windows ones. The OS/2 one was wonderful, but that's water over the dam at this point.)

Don't you find that for any serious use this takes at least an hour to evaluate, each time? I do. I hate random websites because they aren't always trustworthy. There are shining exceptions such as Sam's Laser FAQ.

Delivering a result to the screen isn't the same as delivering it in useful form to the brain. If you live in Mathematica, it's probably an efficient delivery system for you. I just like books.

How would one go about checking such a claim?

I like things that *sit still*. Messing about with software like that makes me feel like Sisyphus--I just get to the top of the hill, and then out comes Rock 3.0 and I'm back where I started. I want it to still be there 20 years from now. I have enough complexity to manage without having my sources be moving targets.

I couldn't disagree more. For light duty stuff that might be true, but the number of insidious gotchas in serious numerical work is mind-boggling. Heavy-duty numerical work is all done on multiprocessors. I've written my own clusterized FDTD simulator because nothing I could buy would do what I needed. I wouldn't ever use an algorithm I didn't understand for any serious purpose. (I don't mean that I have to trawl through the code of, e.g., FFTW, but I know how FFTs work and can check the results easily.) People using e.g. LabView who don't know how it works get into extreme difficulties, because software is so arbitrary.

Welcome to Rock 3.0. That's happened to me too--but never with my own code, or any open source I've used. My sympathies.

Cheers,

Phil Hobbs

A good point. The main reasons I can think of would be for a particular source of very specialised data that nobody has bothered to www so far, and for data tables that give references to the sources of the data, something that can be harder to find on www.

Electronic versions of data tables can be very convenient for practical use, for example if you want to interpolate, or fit a curve to the data - starting with it cut-and-pasteable saves much retyping and errors.

But a paper list of, e.g., thermal properties of a class of similar substances, can be useful for choosing a particular one. Easy to google for a property of a single substance, harder to find good lists for comparison.

Abramowitz & Stegun is public domain, and available on www (not on google books AFAICT, but google and you will find). Thus, cheaper than Mathematica, and takes less shelf space.

Numerical methods are very much an exception to this. Also, the mathematics behind what you want to calculate numerically. I don't mean the mathematics of the numerical methods themselves, but, for example, if you want to calculate some arcane special function quickly, accurately, and reliably as a tiny component of a much larger piece of software, being to look up various identities, recurrence relations, etc can be good.

This doesn't require a book. This kind of info is available on google, on wolfram.com. You suggest that it's possible to get Mathematica to cough it up on demand (although the >50 pages on spherical Bessel functions from functions.wolfram.com might be a bit much).

But for when you just want a number, Mathematica (or often enough Matlab or octave) is fine. But I don't like the cult of precision that can result

- when one will compare a number from simple (and approximate) theory with an experimental result with an error of 3%, you don't need many digits in the theoretical result. And since the theory is approximate anyway (in most cases), 16 digits of precision is downright misleading.

What's essential is knowing when they work and when they fail. Having smart black boxes that issue warnings is nice, and generally more reliable than coding your own. Black boxes that don't issue warnings or errors are not so nice, but are still generally more reliable than writing your own code.

Morrison is one of my favorites too. The second edition was my first introduction to the estimation of parisitic circuits, and it bailed me out enough times I bought the third and fourth editions, but not yet the new "5th edition" which has just come out:

Glen

Hi Phil,

Systems",=20

Shouldn't Oppenheim and Shafer. Digital Signal Processing be in the DSP section?

For Circuits, I would add: Thomas H. Lee, Planar Microwave Engineering: A Practical Guide to Theory, Measurements and Circuits. The reviews at Amazon should give you some feel for this book.

Clarke and Hess, Communications Circuits: Analysis and Design. This covers many of the non-linear circuits including, oscillators, mixers, Amplitude modulators and detectors, Frequency modulators and detectors, and power amps. (I believe Clarke was the first to correctly describe the limiting process in Colpitts oscillators and how to _design_ for a given signal level out.)

You might want to have a section on NOISE. In addition to Van der Ziel you could have: Davenport and Root, Random Signals and Noise A classic.

Two books from IEEE Press that are reprints of various papers, many of them classics: Madhu S. Gupta, ed., Electrical Noise: Fundamentals & Sources Hatsuaki Fukui, ed., Low-Noise Microwave Transistors & Amplifiers

For low frequency e_n / i_n type analysis, Motchenbacher and Connelly, Low-Noise Electronic System Design (1993) (First edition was Motchenbacher and Fitchen, Low-noise Electronic = Design (1973))

--=20 Regards, Howard snipped-for-privacy@ix.netcom.com